InfiniteTalk: Audio-Portrait to Lip-Synced Video in ComfyUI

🚀 Create realistic talking avatar videos from a single portrait and voice input — with accurate lip-sync and identity-stable animation.

▶️ Run Directly in Cloud:
/>

💡 Overview

InfiniteTalk is a ComfyUI workflow that generates lip-synced talking videos from a single image and voice input. Powered by the MultiTalk AI model, it produces fluid, identity-stable portrait clips with natural speech motion and prompt-driven customizable animation.

Ideal for content creators, educators, marketers, and anyone who needs realistic talking avatars without filming.

✨ Key Features

• Single Image + Audio: Just provide a portrait and a voice clip — the workflow handles the rest.

• Accurate Lip-Sync: Natural mouth movements precisely synchronized to the audio input.

• Identity Preservation: Facial structure, expression style, and appearance remain consistent throughout.

• Prompt-Driven Customization: Fine-tune animation behavior and visual style with text prompts.

🚀 Getting Started

1. Upload a portrait image — clear, well-lit, forward-facing works best.

2. Provide an audio clip — speech or narration you want the avatar to speak.

3. Generate — the workflow produces a lip-synced video with the original audio muxed in.

Click the "Run Directly" link above to bypass local setup and test this workflow immediately in your browser.